FIRST-PRINCIPLES MODELLING OF DOPANTS AT INTERFACES IN TCO MATERIALS

Transcription

1 FIRST-PRINCIPLES MODELLING OF DOPANTS AT INTERFACES IN TCO MATERIALS Wolfgang Körner and Christian Elsässer, Fraunhofer Institute for Mechanics of Materials IWM, Freiburg, Germany Workshop Transparente und flexible Elektronik Universität Leipzig, 28. September 2010 ( ) m i x i f i Ĥ E

2 First-principles DFT study of dopant elements at grain boundaries in the TCO materials ZnO and TiO 2 Wolfgang Körner and Christian Elsässer, Fraunhofer IWM 1 m glass TCO a-si- p-i-n Ag 2

3 METCO a Multidisciplinary Effort towards advanced Transparent and Conducting Oxide electrodes B. Szyszka 1, K. Ortner 1, S. Ulrich 1, C. Polenzky 1, V. Sittinger 1, F. Horstmann 1, A. Georg 2, D. Borchert 2, P. Löbmann 3, S. Götzendorfer 3, C. May 4,L. Ruppel 4, C. Elsässer 5, W. Körner 5, E. Kalivoda 5 1) Fraunhofer IST, Braunschweig, Germany 2) Fraunhofer ISE, Freiburg, Germany 3) Fraunhofer ISC, Würzburg, Germany 4) Fraunhofer IPMS, Dresden, Germany 5) Fraunhofer IWM, Freiburg, Germany 3

4 Transparent and conductive oxide thin-film systems need: large-area electrical contacts for solar cells and light emitting diodes h demands: Transp. anode Hole conductor Active layer Metal. cathode + high optical transmittance and electrical conductance proper matching of band structures at interfaces sufficient light scattering for thin-film solar cells capability for patterning and processing low-cost film deposition on large areas desire: p-doping p-n hetero-junctions transparent electronics 4

5 ZnO TCO: transparent and conducting oxide ZnO:X (X = N, P, Al, Ga) alternative material to ITO large-area thin-film top electrode for solar cells or OLED devices polycrystalline microstructure interaction of dopants with grain boundaries DFT: density functional theory LDA: local density approximation, for optimisation of atomic structures formation energies of structural defects SIC: self interaction correction, for analysis of electronic structures defect levels in energy gap of more accurate band structure supercell models, mixed basis, pseudopotentials (MBPP code) 5

6 First-principles Density Functional Theory Mixed-basis pseudopotential method Meyer, Elsässer, Lechermann, Fähnle, et al. (origin: MPI-MF Stuttgart) density functional theory LDA, GGA SIC-LDA, LDA+U translational lattice symmetry core-valence interactions valence electrons periodic supercell models norm-conserving pseudopotentials plane waves and localized orbitals output: energies and forces crystal structures, defect configurations electronic structure, chemical bonding phase stabilities, defect energies 6

7 Band structure of bulk ZnO LDA 1.13 ev expt.: 3.4 ev 7

8 Band structure of bulk ZnO SIC 3.39 ev expt.: 3.4 ev 8 bulk SIC: Zn-3d 100%, O-2s 100%, O-2p 80%, α = 0.8

9 Band structure of bulk ZnO SIC formalism: SIC pseudopotentials cf. Vogel, Krüger, Pollmann Phys. Rev. B 54 (1996) ev expt.: 3.4 ev 9 other SIC PP implementations, e.g.: Filippetti and Spaldin Phys. Rev. B 67 (2003) Pemmaraju et al. Phys. Rev. B 75 (2007) bulk SIC: Zn-3d 100%, O-2s 100%, O-2p 80%, α = 0.8

10 Grain boundaries in ceramic ZnO bicrystal experiments GB3 GB2 10 Sato et al., J. Am. Ceram. Soc. 90 (2007)

11 Supercells for three different GB in ZnO 80 atoms E GB1 = 0.17 J/m atoms E GB2 = 1.89 J/m atoms E GB3 = 1.65 J/m 2 11

12 reference system: band structure of bulk ZnO SIC-LDA Zn 4s O 2p band gap 3.4 ev Zn 3d DOS (1 / ev) valence band conduction band energy (ev) 12

13 reference system: band structure of bulk ZnO SIC-LDA DOS (1 / ev) Zn 3d O 2p valence band Zn 4s conduction band energy (ev) 13

14 LDA: total DOS of undoped bulk and GB 14

15 SIC: total DOS of undoped bulk and GB 15

16 LDA + SIC: formation energies for dopants in bulk ZnO For calculation of defect formation energies see, e.g.: Van de Walle and Neugebauer J. Appl. Phys. 95 (2003) For discussions of O vacancy in bulk ZnO see, e.g.: Lany and Zunger Phys. Rev. B 78 (2008) Van de Walle J. Phys.: Condens. Matter 20 (2008) (Zn vacancy in bulk ZnO is tricky as well.)

17 indicators for conductivity and transparency in band structure valence band conduction band 1. undoped semiconductor DOS (1 / ev) E 1 E 2 2. p-doping (bad) 3. p-doping (good) 4. n-doping (bad) occupied with electrons energy (ev) 5. n-doping (good) many states in the bulk band gap reduced transparency by transitions: hc / λ = E 2 -E 1 17

18 SIC: oxygen vacancy in ZnO O GB2: bad conductivity, and bad transparency exp.: green luminescence" of ZnO with energy 2.4 ev 18

19 SIC: N O dopant in ZnO N GB2: p-type conductivity at room temp. plausible, and good transparency 19

20 SIC: P O dopant in ZnO P GB2: good conductivity, but bad transparency 20

21 SIC: Al Zn dopant in ZnO Al GB2: n-type conductivity at room temp. plausible, and good transparency 21

22 SIC: Ga Zn dopant in ZnO Ga GB2: good conductivity, and good transparency 22

23 Conclusion I: ZnO SIC: self interaction correction via pseudopotentials can describe band structure of ZnO better than LDA. It is applicable to extended structural defects like grain boundaries. grain boundaries in polycrystalline ZnO can cause deep defect levels in the band gap. This SIC result is different from others and our LDA results for GB! dopants at grain boundaries in ZnO can cause shallow defect levels in the band gap. Considering grain boundaries, not only perfect crystals, may help for better understanding of dopants in TCO ceramics? Reference: W. Körner, C. Elsässer, Phys. Rev. B 81 (2010) Funding: Fraunhofer Society 23

24 collection of supercell models for grain boundaries material boundary interface system size γ [J/m^2] Wurtzite ZnO ( 1010)[1010] 1 [1010] / GB1 ( 1230)[0001] 7 [0001] / / /1.88 GB2 ( 2350)[0001] 19 [0001] / GB3 Rutile TiO 2 ( 100)[100] 1 [100] / 180 ( 210)[001] 5 [001] / / /1.92 GB1 GB2 ( 310)[001] 5 [001] / GB3 Anatase TiO 2 ( 100)[100] 1 [100] / GB1 ( 021)[100] 5 ( 031)[100] 5 ( 120)[001] 5 [100] [100] [001] / / / GB2 GB3 GB4 24

25 TiO 2 : band structure of Rutile LDA E gap = 3.04 ev (exp.) SIC Ti 3d O 2p E gap = 1.55 ev E gap = 3.02 ev bulk SIC: Ti (LDA), O-2s 100%, O-2p 90%, α =

26 TiO2: band structure of Anatase LDA E gap = 3.20 ev (exp.) SIC Ti 3d O 2p E gap = 1.73 ev E gap = 3.22 ev bulk SIC: Ti (LDA), O-2s 100%, O-2p 80%, α =

27 DOS of undoped GB in TiO 2 (rutile) SIC-LDA deep levels for GB3 GB1 is structually very similar to bulk O atoms at GB3 have only two Ti atoms as nearest neighbours. 27

28 DOS of undoped GB in TiO 2 (anatase) SIC-LDA deep levels for GB1,GB2 and GB3 O atoms at GB1, GB2, and GB3 have only two Ti atoms as nearest neighbours. 28

29 O vacancy in TiO 2 (rutile) comparison LDA vs SIC no donor levels (in accordance with other DFT studies) 29

30 O vacancy in TiO 2 (anatase) comparison LDA vs SIC no donor levels (in accordance with other DFT studies) 30

31 Interstitial Ti atom in TiO 2 (rutile) comparison LDA vs SIC donor levels exp.: "blue coloration" of reduced TiO 2 (rutile) with energy 0.75eV Cronemeyer et al. Phys. Rev. 82, 975 (1951) 31

32 Interstitial Ti atom in TiO 2 (anatase) comparison LDA vs SIC donor levels Hence: 1. Anatase und Rutile show very similar behaviour. 2. Already LDA yields donor states. 3. SIC yields levels which could explain blue coloration. 32

33 TiO 2 (rutile) doped with Nb Ti comparison bulk vs GB2 Nb causes additional levels below the conduction-band edge. hence: Nb is suitable for n-doping. 33

34 TiO 2 (rutile) doped with P O Vergleich Bulk vs GB2 At GB2 the additional levels are more smeared out than in bulk. nevertheless: P is not suitable for p-doping. 34

35 Conclusion II: ZnO and TiO 2 Undoped grain boundaries of ZnO and TiO 2, which have O atoms with dangling bonds, show deep levels. TiO 2 : Anatase and Rutile show only little differences. Intrinsic point defects O-vacancy in ZnO and Ti-interstitals in TiO 2 may be related to n-conductivity. experimental green luminescence (2.4eV) of ZnO or "blue coloration" (0.75eV) of TiO 2 may be understood from theoretical SIC results. Extrinsic point defects N is a good candidate for p-doping in both ZnO and TiO 2 Al and Ga are good candidates for n-doping in ZnO, Nb is the best found one in TiO 2. 35

36 First-principles DFT study of dopant elements at grain boundaries in the TCO materials ZnO and TiO 2 Wolfgang Körner and Christian Elsässer, Fraunhofer IWM 1 m glass TCO a-si- p-i-n Ag 36

37 Physical modelling of materials at Fraunhofer IWM Matous Mrovec Adham Hashibon Jan Michael Albina Ĥ E Pavel Marton Wolfgang Körner Christian Elsässer 37 Tony Paxton (QUB) Pierre Hirel Sabine Körbel Eva Marie Kalivoda Martin Reese Paul Bristowe (CAM)